import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
%matplotlib inline
import seaborn as sns

Part A - Domain - Telecom

1. Data understanding and Exploration

# 1A. Reading the Churn_1 csv file

churn1_df = pd.read_csv("TelcomCustomer-Churn_1.csv")
churn1_df.head()

	customerID	gender	SeniorCitizen	Partner	Dependents	tenure	PhoneService	MultipleLines	InternetService	OnlineSecurity
0	7590-VHVEG	Female	0	Yes	No	1	No	No phone service	DSL	No
1	5575-GNVDE	Male	0	No	No	34	Yes	No	DSL	Yes
2	3668-QPYBK	Male	0	No	No	2	Yes	No	DSL	Yes
3	7795-CFOCW	Male	0	No	No	45	No	No phone service	DSL	Yes
4	9237-HQITU	Female	0	No	No	2	Yes	No	Fiber optic	No

#1B.  Reading the Churn_2 csv file

churn2_df = pd.read_csv("TelcomCustomer-Churn_2.csv")
churn2_df.head()

	customerID	OnlineBackup	DeviceProtection	TechSupport	StreamingTV	StreamingMovies	Contract	PaperlessBilling	PaymentMethod	MonthlyCharges	TotalCharges	Churn
0	7590-VHVEG	Yes	No	No	No	No	Month-to-month	Yes	Electronic check	29.85	29.85	No
1	5575-GNVDE	No	Yes	No	No	No	One year	No	Mailed check	56.95	1889.5	No
2	3668-QPYBK	Yes	No	No	No	No	Month-to-month	Yes	Mailed check	53.85	108.15	Yes
3	7795-CFOCW	No	Yes	Yes	No	No	One year	No	Bank transfer (automatic)	42.30	1840.75	No
4	9237-HQITU	No	No	No	No	No	Month-to-month	Yes	Electronic check	70.70	151.65	Yes

print("Churn 1",churn1_df.count())
print("")
print("Churn 2",churn2_df.count())

Churn 1 customerID         7043
gender             7043
SeniorCitizen      7043
Partner            7043
Dependents         7043
tenure             7043
PhoneService       7043
MultipleLines      7043
InternetService    7043
OnlineSecurity     7043
dtype: int64

Churn 2 customerID          7043
OnlineBackup        7043
DeviceProtection    7043
TechSupport         7043
StreamingTV         7043
StreamingMovies     7043
Contract            7043
PaperlessBilling    7043
PaymentMethod       7043
MonthlyCharges      7043
TotalCharges        7043
Churn               7043
dtype: int64

#1C.... Both the dataframes have equal number of rows 7043 and they have to be merged horizontally on the Customer ID

comb_df1 = pd.merge(churn1_df, churn2_df, on='customerID', indicator=True)

comb_df1   #............... _merge column indicates both the dataframes are merged

	customerID	gender	SeniorCitizen	Partner	Dependents	tenure	PhoneService	MultipleLines	InternetService	OnlineSecurity	...	TechSupport	StreamingTV	StreamingMovies	Contract	PaperlessBilling	PaymentMethod	MonthlyCharges	TotalCharges	Churn	_merge
0	7590-VHVEG	Female	0	Yes	No	1	No	No phone service	DSL	No	...	No	No	No	Month-to-month	Yes	Electronic check	29.85	29.85	No	both
1	5575-GNVDE	Male	0	No	No	34	Yes	No	DSL	Yes	...	No	No	No	One year	No	Mailed check	56.95	1889.5	No	both
2	3668-QPYBK	Male	0	No	No	2	Yes	No	DSL	Yes	...	No	No	No	Month-to-month	Yes	Mailed check	53.85	108.15	Yes	both
3	7795-CFOCW	Male	0	No	No	45	No	No phone service	DSL	Yes	...	Yes	No	No	One year	No	Bank transfer (automatic)	42.30	1840.75	No	both
4	9237-HQITU	Female	0	No	No	2	Yes	No	Fiber optic	No	...	No	No	No	Month-to-month	Yes	Electronic check	70.70	151.65	Yes	both
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
7038	6840-RESVB	Male	0	Yes	Yes	24	Yes	Yes	DSL	Yes	...	Yes	Yes	Yes	One year	Yes	Mailed check	84.80	1990.5	No	both
7039	2234-XADUH	Female	0	Yes	Yes	72	Yes	Yes	Fiber optic	No	...	No	Yes	Yes	One year	Yes	Credit card (automatic)	103.20	7362.9	No	both
7040	4801-JZAZL	Female	0	Yes	Yes	11	No	No phone service	DSL	Yes	...	No	No	No	Month-to-month	Yes	Electronic check	29.60	346.45	No	both
7041	8361-LTMKD	Male	1	Yes	No	4	Yes	Yes	Fiber optic	No	...	No	No	No	Month-to-month	Yes	Mailed check	74.40	306.6	Yes	both
7042	3186-AJIEK	Male	0	No	No	66	Yes	No	Fiber optic	Yes	...	Yes	Yes	Yes	Two year	Yes	Bank transfer (automatic)	105.65	6844.5	No	both

7043 rows × 22 columns

comb_df = pd.merge(churn1_df, churn2_df, on='customerID')

#1D.  All the columns have been added to the new dataframes

print("churn1 shape",churn1_df.shape)
print("churn2 shape",churn2_df.shape)
print("combined df",comb_df.shape)

churn1 shape (7043, 10)
churn2 shape (7043, 12)
combined df (7043, 21)

churn1_df.eq(churn2_df).all(axis=1)

0       False
1       False
2       False
3       False
4       False
        ...  
7038    False
7039    False
7040    False
7041    False
7042    False
Length: 7043, dtype: bool

churn1_df.columns

Index(['customerID', 'gender', 'SeniorCitizen', 'Partner', 'Dependents',
       'tenure', 'PhoneService', 'MultipleLines', 'InternetService',
       'OnlineSecurity'],
      dtype='object')

churn2_df.columns

Index(['customerID', 'OnlineBackup', 'DeviceProtection', 'TechSupport',
       'StreamingTV', 'StreamingMovies', 'Contract', 'PaperlessBilling',
       'PaymentMethod', 'MonthlyCharges', 'TotalCharges', 'Churn'],
      dtype='object')

comb_df.columns

Index(['customerID', 'gender', 'SeniorCitizen', 'Partner', 'Dependents',
       'tenure', 'PhoneService', 'MultipleLines', 'InternetService',
       'OnlineSecurity', 'OnlineBackup', 'DeviceProtection', 'TechSupport',
       'StreamingTV', 'StreamingMovies', 'Contract', 'PaperlessBilling',
       'PaymentMethod', 'MonthlyCharges', 'TotalCharges', 'Churn'],
      dtype='object')

comb_df.dtypes

customerID           object
gender               object
SeniorCitizen         int64
Partner              object
Dependents           object
tenure                int64
PhoneService         object
MultipleLines        object
InternetService      object
OnlineSecurity       object
OnlineBackup         object
DeviceProtection     object
TechSupport          object
StreamingTV          object
StreamingMovies      object
Contract             object
PaperlessBilling     object
PaymentMethod        object
MonthlyCharges      float64
TotalCharges         object
Churn                object
dtype: object

comb_df.describe()

	SeniorCitizen	tenure	MonthlyCharges
count	7043.000000	7043.000000	7043.000000
mean	0.162147	32.371149	64.761692
std	0.368612	24.559481	30.090047
min	0.000000	0.000000	18.250000
25%	0.000000	9.000000	35.500000
50%	0.000000	29.000000	70.350000
75%	0.000000	55.000000	89.850000
max	1.000000	72.000000	118.750000

comb_df.head()

	customerID	gender	SeniorCitizen	Partner	Dependents	tenure	PhoneService	MultipleLines	InternetService	OnlineSecurity	...	DeviceProtection	TechSupport	StreamingTV	StreamingMovies	Contract	PaperlessBilling	PaymentMethod	MonthlyCharges	TotalCharges	Churn
0	7590-VHVEG	Female	0	Yes	No	1	No	No phone service	DSL	No	...	No	No	No	No	Month-to-month	Yes	Electronic check	29.85	29.85	No
1	5575-GNVDE	Male	0	No	No	34	Yes	No	DSL	Yes	...	Yes	No	No	No	One year	No	Mailed check	56.95	1889.5	No
2	3668-QPYBK	Male	0	No	No	2	Yes	No	DSL	Yes	...	No	No	No	No	Month-to-month	Yes	Mailed check	53.85	108.15	Yes
3	7795-CFOCW	Male	0	No	No	45	No	No phone service	DSL	Yes	...	Yes	Yes	No	No	One year	No	Bank transfer (automatic)	42.30	1840.75	No
4	9237-HQITU	Female	0	No	No	2	Yes	No	Fiber optic	No	...	No	No	No	No	Month-to-month	Yes	Electronic check	70.70	151.65	Yes

5 rows × 21 columns

2. Data clearning and Analysis

#2A.... Impute missing/unexpected values....

comb_df.isnull().sum()

customerID          0
gender              0
SeniorCitizen       0
Partner             0
Dependents          0
tenure              0
PhoneService        0
MultipleLines       0
InternetService     0
OnlineSecurity      0
OnlineBackup        0
DeviceProtection    0
TechSupport         0
StreamingTV         0
StreamingMovies     0
Contract            0
PaperlessBilling    0
PaymentMethod       0
MonthlyCharges      0
TotalCharges        0
Churn               0
dtype: int64

comb_df.TotalCharges = pd.to_numeric(comb_df.TotalCharges, errors="coerce")

comb_df.dtypes

customerID           object
gender               object
SeniorCitizen         int64
Partner              object
Dependents           object
tenure                int64
PhoneService         object
MultipleLines        object
InternetService      object
OnlineSecurity       object
OnlineBackup         object
DeviceProtection     object
TechSupport          object
StreamingTV          object
StreamingMovies      object
Contract             object
PaperlessBilling     object
PaymentMethod        object
MonthlyCharges      float64
TotalCharges        float64
Churn                object
dtype: object

comb_df.describe()

	SeniorCitizen	tenure	MonthlyCharges	TotalCharges
count	7043.000000	7043.000000	7043.000000	7032.000000
mean	0.162147	32.371149	64.761692	2283.300441
std	0.368612	24.559481	30.090047	2266.771362
min	0.000000	0.000000	18.250000	18.800000
25%	0.000000	9.000000	35.500000	401.450000
50%	0.000000	29.000000	70.350000	1397.475000
75%	0.000000	55.000000	89.850000	3794.737500
max	1.000000	72.000000	118.750000	8684.800000

comb_df['TotalCharges'].isnull().sum() #......... there are 11 missing values in TotalCharges

11

comb_df.isnull().values.any()

True

comb_df.isnull().sum().sum()

11

df1 = comb_df[comb_df.isna().any(axis=1)]

df1 #........... these are the rows with missing TotalCharges values

	customerID	gender	SeniorCitizen	Partner	Dependents	tenure	PhoneService	MultipleLines	InternetService	OnlineSecurity	...	DeviceProtection	TechSupport	StreamingTV	StreamingMovies	Contract	PaperlessBilling	PaymentMethod	MonthlyCharges	TotalCharges	Churn
488	4472-LVYGI	Female	0	Yes	Yes	0	No	No phone service	DSL	Yes	...	Yes	Yes	Yes	No	Two year	Yes	Bank transfer (automatic)	52.55	NaN	No
753	3115-CZMZD	Male	0	No	Yes	0	Yes	No	No	No internet service	...	No internet service	No internet service	No internet service	No internet service	Two year	No	Mailed check	20.25	NaN	No
936	5709-LVOEQ	Female	0	Yes	Yes	0	Yes	No	DSL	Yes	...	Yes	No	Yes	Yes	Two year	No	Mailed check	80.85	NaN	No
1082	4367-NUYAO	Male	0	Yes	Yes	0	Yes	Yes	No	No internet service	...	No internet service	No internet service	No internet service	No internet service	Two year	No	Mailed check	25.75	NaN	No
1340	1371-DWPAZ	Female	0	Yes	Yes	0	No	No phone service	DSL	Yes	...	Yes	Yes	Yes	No	Two year	No	Credit card (automatic)	56.05	NaN	No
3331	7644-OMVMY	Male	0	Yes	Yes	0	Yes	No	No	No internet service	...	No internet service	No internet service	No internet service	No internet service	Two year	No	Mailed check	19.85	NaN	No
3826	3213-VVOLG	Male	0	Yes	Yes	0	Yes	Yes	No	No internet service	...	No internet service	No internet service	No internet service	No internet service	Two year	No	Mailed check	25.35	NaN	No
4380	2520-SGTTA	Female	0	Yes	Yes	0	Yes	No	No	No internet service	...	No internet service	No internet service	No internet service	No internet service	Two year	No	Mailed check	20.00	NaN	No
5218	2923-ARZLG	Male	0	Yes	Yes	0	Yes	No	No	No internet service	...	No internet service	No internet service	No internet service	No internet service	One year	Yes	Mailed check	19.70	NaN	No
6670	4075-WKNIU	Female	0	Yes	Yes	0	Yes	Yes	DSL	No	...	Yes	Yes	Yes	No	Two year	No	Mailed check	73.35	NaN	No
6754	2775-SEFEE	Male	0	No	Yes	0	Yes	Yes	DSL	Yes	...	No	Yes	No	No	Two year	Yes	Bank transfer (automatic)	61.90	NaN	No

11 rows × 21 columns

comb_df.sort_values(by="TotalCharges")

	customerID	gender	SeniorCitizen	Partner	Dependents	tenure	PhoneService	MultipleLines	InternetService	OnlineSecurity	...	DeviceProtection	TechSupport	StreamingTV	StreamingMovies	Contract	PaperlessBilling	PaymentMethod	MonthlyCharges	TotalCharges	Churn
1654	2967-MXRAV	Male	0	Yes	Yes	1	Yes	No	No	No internet service	...	No internet service	No internet service	No internet service	No internet service	One year	No	Mailed check	18.80	18.80	No
6489	9318-NKNFC	Male	0	No	No	1	Yes	No	No	No internet service	...	No internet service	No internet service	No internet service	No internet service	Month-to-month	Yes	Mailed check	18.85	18.85	Yes
1151	8992-CEUEN	Female	0	No	No	1	Yes	No	No	No internet service	...	No internet service	No internet service	No internet service	No internet service	Month-to-month	No	Electronic check	18.85	18.85	No
4939	9975-SKRNR	Male	0	No	No	1	Yes	No	No	No internet service	...	No internet service	No internet service	No internet service	No internet service	Month-to-month	No	Mailed check	18.90	18.90	No
583	1423-BMPBQ	Female	0	Yes	Yes	1	Yes	No	No	No internet service	...	No internet service	No internet service	No internet service	No internet service	Month-to-month	Yes	Mailed check	19.00	19.00	No
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
3826	3213-VVOLG	Male	0	Yes	Yes	0	Yes	Yes	No	No internet service	...	No internet service	No internet service	No internet service	No internet service	Two year	No	Mailed check	25.35	NaN	No
4380	2520-SGTTA	Female	0	Yes	Yes	0	Yes	No	No	No internet service	...	No internet service	No internet service	No internet service	No internet service	Two year	No	Mailed check	20.00	NaN	No
5218	2923-ARZLG	Male	0	Yes	Yes	0	Yes	No	No	No internet service	...	No internet service	No internet service	No internet service	No internet service	One year	Yes	Mailed check	19.70	NaN	No
6670	4075-WKNIU	Female	0	Yes	Yes	0	Yes	Yes	DSL	No	...	Yes	Yes	Yes	No	Two year	No	Mailed check	73.35	NaN	No
6754	2775-SEFEE	Male	0	No	Yes	0	Yes	Yes	DSL	Yes	...	No	Yes	No	No	Two year	Yes	Bank transfer (automatic)	61.90	NaN	No

7043 rows × 21 columns

# Replace using median 
median = comb_df['TotalCharges'].median()
comb_df['TotalCharges'].fillna(median, inplace=True)

comb_df['TotalCharges'].isnull().sum()  #........... after replacing with median........ there are no more missing values

0

comb_df.describe()

	SeniorCitizen	tenure	MonthlyCharges	TotalCharges
count	7043.000000	7043.000000	7043.000000	7043.000000
mean	0.162147	32.371149	64.761692	2281.916928
std	0.368612	24.559481	30.090047	2265.270398
min	0.000000	0.000000	18.250000	18.800000
25%	0.000000	9.000000	35.500000	402.225000
50%	0.000000	29.000000	70.350000	1397.475000
75%	0.000000	55.000000	89.850000	3786.600000
max	1.000000	72.000000	118.750000	8684.800000

#2A........... Other than TotalCharges column all other columns had no missing values or null values....

# But those Null values in Total Charges were not being picked up by pandas initially and was able to find it after
# the column was changed to_numeric

#... There were 11 Null values in TotalCharges column and they have been replaced by Median values.

for i in comb_df.columns:
    if ((i in ["customerID","tenure","MonthlyCharges","TotalCharges"])):
         print()
    else:   
        print("% S unique values", i)
        print(comb_df[i].value_counts())
        print("")
    

% S unique values gender
Male      3555
Female    3488
Name: gender, dtype: int64

% S unique values SeniorCitizen
0    5901
1    1142
Name: SeniorCitizen, dtype: int64

% S unique values Partner
No     3641
Yes    3402
Name: Partner, dtype: int64

% S unique values Dependents
No     4933
Yes    2110
Name: Dependents, dtype: int64


% S unique values PhoneService
Yes    6361
No      682
Name: PhoneService, dtype: int64

% S unique values MultipleLines
No                  3390
Yes                 2971
No phone service     682
Name: MultipleLines, dtype: int64

% S unique values InternetService
Fiber optic    3096
DSL            2421
No             1526
Name: InternetService, dtype: int64

% S unique values OnlineSecurity
No                     3498
Yes                    2019
No internet service    1526
Name: OnlineSecurity, dtype: int64

% S unique values OnlineBackup
No                     3088
Yes                    2429
No internet service    1526
Name: OnlineBackup, dtype: int64

% S unique values DeviceProtection
No                     3095
Yes                    2422
No internet service    1526
Name: DeviceProtection, dtype: int64

% S unique values TechSupport
No                     3473
Yes                    2044
No internet service    1526
Name: TechSupport, dtype: int64

% S unique values StreamingTV
No                     2810
Yes                    2707
No internet service    1526
Name: StreamingTV, dtype: int64

% S unique values StreamingMovies
No                     2785
Yes                    2732
No internet service    1526
Name: StreamingMovies, dtype: int64

% S unique values Contract
Month-to-month    3875
Two year          1695
One year          1473
Name: Contract, dtype: int64

% S unique values PaperlessBilling
Yes    4171
No     2872
Name: PaperlessBilling, dtype: int64

% S unique values PaymentMethod
Electronic check             2365
Mailed check                 1612
Bank transfer (automatic)    1544
Credit card (automatic)      1522
Name: PaymentMethod, dtype: int64



% S unique values Churn
No     5174
Yes    1869
Name: Churn, dtype: int64

for i in comb_df.columns:
    if ((i in ["customerID","tenure","MonthlyCharges","TotalCharges"])):
        print()
    else:   
        fig = plt.figure(figsize = (10,5))
        plt.rcParams.update({'figure.max_open_warning': 0})
        ax = fig.gca()
        fig.suptitle(i, fontsize=20)
        #plt.figure(i)
        sns.histplot(comb_df[i], kde=True, ax=ax)
        #print("")

comb_df["TotalCharges"].unique()

array([  29.85, 1889.5 ,  108.15, ...,  346.45,  306.6 , 6844.5 ])

#2B..... Converting Continuous variable column TotalCharges to float

comb_df['TotalCharges'] = pd.to_numeric(comb_df['TotalCharges'],errors='coerce')
                                
comb_df['TotalCharges'] = comb_df['TotalCharges'].astype(float)

comb_df.dtypes

customerID           object
gender               object
SeniorCitizen         int64
Partner              object
Dependents           object
tenure                int64
PhoneService         object
MultipleLines        object
InternetService      object
OnlineSecurity       object
OnlineBackup         object
DeviceProtection     object
TechSupport          object
StreamingTV          object
StreamingMovies      object
Contract             object
PaperlessBilling     object
PaymentMethod        object
MonthlyCharges      float64
TotalCharges        float64
Churn                object
dtype: object

# 2C........... Function to accept dataframe and then print pie charts for all Categorical columns

def pie_df(df):
        
    for i in df.columns:
        if (df[i].dtypes == "object"):
            
            if i in ["customerID","TotalCharges"]:
                print()
            else:
                #print("column is %S",i)                
                                
                df.groupby(i).size().plot(kind='pie', subplots=True, shadow=True, startangle=30, figsize=(8,6), autopct='%1.2f%%')                
                font1 = {'family':'serif','color':'blue','size':20}
                plt.title(i, fontdict = font1)    
                
                plt.tight_layout()
                plt.show()

   

pie_df(comb_df)

2D.......... Insights on question 2C

There are totally 19 categorical value columns

1. Gender - There is almost an equi balance of Male and Female customers
2. Partner - There is again almost equi balance of Customers having partners and no partners
3. Dependents - only 1/3rd of the Customers almost 30% have dependents while the remaining 70% have no dependents
4. PhoneService - A huge 90% of the Customers have availed for Phoneservice while a small 10% have not opted for it
5. Multiplelines - About 42% of the customers have multiple lines while 48% of the customers do nto have multiple lines
6. Internetservice - A majority of Customers 44% have Fibre optic, 34% have DSL while 21% have no Internet servicce
7. OnlineSecurity - 50% of customers have not availed Onlinesecurity and only 28% of the customers have taken this service,

                21% in any case have no Internetservice

8. OnlineBackup - Only 34.5% of the Customers have availed this service, 44% have not while 21% have no Internetservice at all
9. DeviceProtection - Numbers almost look similar to the OnlineBackup 34.5 - yes, 44% no, 21% no Internetservice
10. Techsupport - 29% of the customers have only signed up for this, 49% have'nt, while 21% have no Internetservice
11. StreamingTV - Very equally distributed... 38% signed up, 39% not signed up, 21% no Internetservice
12. STreamingMovies - Almost similar distribution as found in StreamingTV
13. Contract - 55% of the customers going for month to month, 21% one year & 24% for two year contracts
14. Paperless Billing - Almost 60% have opted for this, while the remaining 40% are for paper billing
15. PaymentMethod - Credit card choice - 21.6%, Bank transfers - 22%, Electronic transfer - 35%, Mailed checks - 23%
16. Churn - This is the Target variable - 73.5% look to be Loyal customers while 26.5% Churn out

# 2E... Encoding all the Categorical variables 

#classifying the categorical variables into 2 variants (Label encoding & One Hot encoding)

replaceStruct = {
                "MultipleLines":     {"Yes": 1, "No": 0, "No phone service": -1},
                "OnlineSecurity":    {"Yes": 1, "No": 0 ,"No internet service": -1},
                "OnlineBackup":      {"Yes": 1, "No": 0 ,"No internet service": -1},
                "DeviceProtection":  {"Yes": 1, "No": 0 ,"No internet service": -1},
                "TechSupport":       {"Yes": 1, "No": 0 ,"No internet service": -1},
                "StreamingTV":       {"Yes": 1, "No": 0 ,"No internet service": -1},
                "StreamingMovies":   {"Yes": 1, "No": 0 ,"No internet service": -1},
                "Contract":          {"Month-to-month": 1, "One year": 2 ,"Two year": 3},
                "Churn":             {"Yes": 1, "No": 0 },                
                "Partner":           {"Yes": 1, "No": 0 },
                "Dependents":        {"Yes": 1, "No": 0 },
                "PhoneService":      {"Yes": 1, "No": 0 },
                "PaperlessBilling":  {"Yes": 1, "No": 0 } 
                    }
oneHotCols=["PaymentMethod","InternetService","gender"]

replaceStruct

{'MultipleLines': {'Yes': 1, 'No': 0, 'No phone service': -1},
 'OnlineSecurity': {'Yes': 1, 'No': 0, 'No internet service': -1},
 'OnlineBackup': {'Yes': 1, 'No': 0, 'No internet service': -1},
 'DeviceProtection': {'Yes': 1, 'No': 0, 'No internet service': -1},
 'TechSupport': {'Yes': 1, 'No': 0, 'No internet service': -1},
 'StreamingTV': {'Yes': 1, 'No': 0, 'No internet service': -1},
 'StreamingMovies': {'Yes': 1, 'No': 0, 'No internet service': -1},
 'Contract': {'Month-to-month': 1, 'One year': 2, 'Two year': 3},
 'Churn': {'Yes': 1, 'No': 0},
 'Partner': {'Yes': 1, 'No': 0},
 'Dependents': {'Yes': 1, 'No': 0},
 'PhoneService': {'Yes': 1, 'No': 0},
 'PaperlessBilling': {'Yes': 1, 'No': 0}}

oneHotCols

['PaymentMethod', 'InternetService', 'gender']

new_df=comb_df.replace(replaceStruct)
new_df=pd.get_dummies(new_df, columns=oneHotCols)
new_df.head(10)

	customerID	SeniorCitizen	Partner	Dependents	tenure	PhoneService	MultipleLines	OnlineSecurity	OnlineBackup	DeviceProtection	...	Churn	PaymentMethod_Bank transfer (automatic)	PaymentMethod_Credit card (automatic)	PaymentMethod_Electronic check	PaymentMethod_Mailed check	InternetService_DSL	InternetService_Fiber optic	InternetService_No	gender_Female	gender_Male
0	7590-VHVEG	0	1	0	1	0	-1	0	1	0	...	0	0	0	1	0	1	0	0	1	0
1	5575-GNVDE	0	0	0	34	1	0	1	0	1	...	0	0	0	0	1	1	0	0	0	1
2	3668-QPYBK	0	0	0	2	1	0	1	1	0	...	1	0	0	0	1	1	0	0	0	1
3	7795-CFOCW	0	0	0	45	0	-1	1	0	1	...	0	1	0	0	0	1	0	0	0	1
4	9237-HQITU	0	0	0	2	1	0	0	0	0	...	1	0	0	1	0	0	1	0	1	0
5	9305-CDSKC	0	0	0	8	1	1	0	0	1	...	1	0	0	1	0	0	1	0	1	0
6	1452-KIOVK	0	0	1	22	1	1	0	1	0	...	0	0	1	0	0	0	1	0	0	1
7	6713-OKOMC	0	0	0	10	0	-1	1	0	0	...	0	0	0	0	1	1	0	0	1	0
8	7892-POOKP	0	1	0	28	1	1	0	0	1	...	1	0	0	1	0	0	1	0	1	0
9	6388-TABGU	0	0	1	62	1	0	1	1	0	...	0	1	0	0	0	1	0	0	0	1

10 rows × 27 columns

for i in new_df.columns:
    if ((i in ["customerID","tenure","MonthlyCharges","TotalCharges"])):
         print()
    else:   
        print("% S unique values", i)
        print(new_df[i].value_counts())
        print("")
    

% S unique values SeniorCitizen
0    5901
1    1142
Name: SeniorCitizen, dtype: int64

% S unique values Partner
0    3641
1    3402
Name: Partner, dtype: int64

% S unique values Dependents
0    4933
1    2110
Name: Dependents, dtype: int64


% S unique values PhoneService
1    6361
0     682
Name: PhoneService, dtype: int64

% S unique values MultipleLines
 0    3390
 1    2971
-1     682
Name: MultipleLines, dtype: int64

% S unique values OnlineSecurity
 0    3498
 1    2019
-1    1526
Name: OnlineSecurity, dtype: int64

% S unique values OnlineBackup
 0    3088
 1    2429
-1    1526
Name: OnlineBackup, dtype: int64

% S unique values DeviceProtection
 0    3095
 1    2422
-1    1526
Name: DeviceProtection, dtype: int64

% S unique values TechSupport
 0    3473
 1    2044
-1    1526
Name: TechSupport, dtype: int64

% S unique values StreamingTV
 0    2810
 1    2707
-1    1526
Name: StreamingTV, dtype: int64

% S unique values StreamingMovies
 0    2785
 1    2732
-1    1526
Name: StreamingMovies, dtype: int64

% S unique values Contract
1    3875
3    1695
2    1473
Name: Contract, dtype: int64

% S unique values PaperlessBilling
1    4171
0    2872
Name: PaperlessBilling, dtype: int64



% S unique values Churn
0    5174
1    1869
Name: Churn, dtype: int64

% S unique values PaymentMethod_Bank transfer (automatic)
0    5499
1    1544
Name: PaymentMethod_Bank transfer (automatic), dtype: int64

% S unique values PaymentMethod_Credit card (automatic)
0    5521
1    1522
Name: PaymentMethod_Credit card (automatic), dtype: int64

% S unique values PaymentMethod_Electronic check
0    4678
1    2365
Name: PaymentMethod_Electronic check, dtype: int64

% S unique values PaymentMethod_Mailed check
0    5431
1    1612
Name: PaymentMethod_Mailed check, dtype: int64

% S unique values InternetService_DSL
0    4622
1    2421
Name: InternetService_DSL, dtype: int64

% S unique values InternetService_Fiber optic
0    3947
1    3096
Name: InternetService_Fiber optic, dtype: int64

% S unique values InternetService_No
0    5517
1    1526
Name: InternetService_No, dtype: int64

% S unique values gender_Female
0    3555
1    3488
Name: gender_Female, dtype: int64

% S unique values gender_Male
1    3555
0    3488
Name: gender_Male, dtype: int64

new_df

	customerID	SeniorCitizen	Partner	Dependents	tenure	PhoneService	MultipleLines	OnlineSecurity	OnlineBackup	DeviceProtection	...	Churn	PaymentMethod_Bank transfer (automatic)	PaymentMethod_Credit card (automatic)	PaymentMethod_Electronic check	PaymentMethod_Mailed check	InternetService_DSL	InternetService_Fiber optic	InternetService_No	gender_Female	gender_Male
0	7590-VHVEG	0	1	0	1	0	-1	0	1	0	...	0	0	0	1	0	1	0	0	1	0
1	5575-GNVDE	0	0	0	34	1	0	1	0	1	...	0	0	0	0	1	1	0	0	0	1
2	3668-QPYBK	0	0	0	2	1	0	1	1	0	...	1	0	0	0	1	1	0	0	0	1
3	7795-CFOCW	0	0	0	45	0	-1	1	0	1	...	0	1	0	0	0	1	0	0	0	1
4	9237-HQITU	0	0	0	2	1	0	0	0	0	...	1	0	0	1	0	0	1	0	1	0
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
7038	6840-RESVB	0	1	1	24	1	1	1	0	1	...	0	0	0	0	1	1	0	0	0	1
7039	2234-XADUH	0	1	1	72	1	1	0	1	1	...	0	0	1	0	0	0	1	0	1	0
7040	4801-JZAZL	0	1	1	11	0	-1	1	0	0	...	0	0	0	1	0	1	0	0	1	0
7041	8361-LTMKD	1	1	0	4	1	1	0	0	0	...	1	0	0	0	1	0	1	0	0	1
7042	3186-AJIEK	0	0	0	66	1	0	1	0	1	...	0	1	0	0	0	0	1	0	0	1

7043 rows × 27 columns

new_df.shape

(7043, 27)

# From 21 columns it has increased to 27 since there are new columns added for Payment Method, Internet service & gender
# mentioned in One Hot encoding

new_df.corr()

	SeniorCitizen	Partner	Dependents	tenure	PhoneService	MultipleLines	OnlineSecurity	OnlineBackup	DeviceProtection	TechSupport	...	Churn	PaymentMethod_Bank transfer (automatic)	PaymentMethod_Credit card (automatic)	PaymentMethod_Electronic check	PaymentMethod_Mailed check	InternetService_DSL	InternetService_Fiber optic	InternetService_No	gender_Female	gender_Male
SeniorCitizen	1.000000	0.016479	-0.211185	0.016567	0.008576	0.113791	0.081878	0.144828	0.140298	0.067457	...	0.150889	-0.016159	-0.024135	0.171718	-0.153477	-0.108322	0.255338	-0.182742	0.001874	-0.001874
Partner	0.016479	1.000000	0.452676	0.379697	0.017706	0.117307	0.091303	0.090753	0.098664	0.076551	...	-0.150448	0.110706	0.082029	-0.083852	-0.095125	-0.000851	0.000304	0.000615	0.001808	-0.001808
Dependents	-0.211185	0.452676	1.000000	0.159712	-0.001762	-0.019657	-0.029721	-0.062775	-0.069078	-0.040788	...	-0.164221	0.052021	0.060267	-0.150642	0.059071	0.052010	-0.165818	0.139812	-0.010517	0.010517
tenure	0.016567	0.379697	0.159712	1.000000	0.008448	0.258958	0.232373	0.253743	0.253999	0.230522	...	-0.352229	0.243510	0.233006	-0.208363	-0.233852	0.013274	0.019720	-0.039062	-0.005106	0.005106
PhoneService	0.008576	0.017706	-0.001762	0.008448	1.000000	0.675070	-0.159989	-0.129770	-0.142012	-0.161930	...	0.011942	0.007556	-0.007721	0.003062	-0.003319	-0.452425	0.289999	0.172209	0.006488	-0.006488
MultipleLines	0.113791	0.117307	-0.019657	0.258958	0.675070	1.000000	0.069085	0.130619	0.124502	0.069149	...	0.036310	0.061513	0.042590	0.065663	-0.176117	-0.361806	0.414749	-0.082560	0.009451	-0.009451
OnlineSecurity	0.081878	0.091303	-0.029721	0.232373	-0.159989	0.069085	1.000000	0.705166	0.701690	0.735191	...	0.023309	0.062184	0.073520	0.094304	-0.239277	0.427891	0.252145	-0.797084	0.014418	-0.014418
OnlineBackup	0.144828	0.090753	-0.062775	0.253743	-0.129770	0.130619	0.705166	1.000000	0.711335	0.709216	...	0.074205	0.057192	0.057873	0.158719	-0.291444	0.314037	0.366555	-0.803663	0.012230	-0.012230
DeviceProtection	0.140298	0.098664	-0.069078	0.253999	-0.142012	0.124502	0.701690	0.711335	1.000000	0.725327	...	0.084654	0.054689	0.071243	0.156932	-0.300069	0.306717	0.373425	-0.803500	0.004720	-0.004720
TechSupport	0.067457	0.076551	-0.040788	0.230522	-0.161930	0.069149	0.735191	0.709216	0.725327	1.000000	...	0.027037	0.066123	0.074562	0.092152	-0.241759	0.422120	0.257847	-0.797300	0.009409	-0.009409
StreamingTV	0.167188	0.079785	-0.086739	0.201053	-0.108239	0.165254	0.662220	0.703167	0.748182	0.703866	...	0.164673	0.030875	0.025425	0.248025	-0.334131	0.217634	0.465190	-0.811354	0.008673	-0.008673
StreamingMovies	0.176469	0.075098	-0.101475	0.205031	-0.114715	0.162841	0.667033	0.700158	0.752952	0.704587	...	0.163220	0.032405	0.030648	0.243398	-0.335553	0.223262	0.460472	-0.812158	0.010011	-0.010011
Contract	-0.142554	0.294806	0.243187	0.671607	0.002247	0.083343	0.015824	-0.035407	0.005848	0.047218	...	-0.396713	0.186440	0.210659	-0.342575	-0.004882	0.055352	-0.254157	0.242388	-0.000126	0.000126
PaperlessBilling	0.156530	-0.014877	-0.111377	0.006152	0.016505	0.133255	0.184993	0.260715	0.246069	0.211031	...	0.191825	-0.016332	-0.013589	0.208865	-0.205398	-0.063121	0.326853	-0.321013	0.011754	-0.011754
MonthlyCharges	0.220173	0.096848	-0.113890	0.247900	0.247398	0.490700	0.635534	0.710477	0.737104	0.661032	...	0.193356	0.042812	0.030550	0.271625	-0.377437	-0.160189	0.787066	-0.763557	0.014569	-0.014569
TotalCharges	0.102652	0.318364	0.063593	0.825464	0.113013	0.412166	0.482605	0.537169	0.545529	0.495075	...	-0.199037	0.186025	0.182745	-0.059971	-0.294814	-0.052279	0.361045	-0.374706	0.000002	-0.000002
Churn	0.150889	-0.150448	-0.164221	-0.352229	0.011942	0.036310	0.023309	0.074205	0.084654	0.027037	...	1.000000	-0.117937	-0.134302	0.301919	-0.091683	-0.124214	0.308020	-0.227890	0.008612	-0.008612
PaymentMethod_Bank transfer (automatic)	-0.016159	0.110706	0.052021	0.243510	0.007556	0.061513	0.062184	0.057192	0.054689	0.066123	...	-0.117937	1.000000	-0.278215	-0.376762	-0.288685	0.025476	-0.022624	-0.002113	0.016024	-0.016024
PaymentMethod_Credit card (automatic)	-0.024135	0.082029	0.060267	0.233006	-0.007721	0.042590	0.073520	0.057873	0.071243	0.074562	...	-0.134302	-0.278215	1.000000	-0.373322	-0.286049	0.051438	-0.050077	0.001030	-0.001215	0.001215
PaymentMethod_Electronic check	0.171718	-0.083852	-0.150642	-0.208363	0.003062	0.065663	0.094304	0.158719	0.156932	0.092152	...	0.301919	-0.376762	-0.373322	1.000000	-0.387372	-0.104418	0.336410	-0.284917	-0.000752	0.000752
PaymentMethod_Mailed check	-0.153477	-0.095125	0.059071	-0.233852	-0.003319	-0.176117	-0.239277	-0.291444	-0.300069	-0.241759	...	-0.091683	-0.288685	-0.286049	-0.387372	1.000000	0.041899	-0.306834	0.321361	-0.013744	0.013744
InternetService_DSL	-0.108322	-0.000851	0.052010	0.013274	-0.452425	-0.361806	0.427891	0.314037	0.306717	0.422120	...	-0.124214	0.025476	0.051438	-0.104418	0.041899	1.000000	-0.640987	-0.380635	-0.006568	0.006568
InternetService_Fiber optic	0.255338	0.000304	-0.165818	0.019720	0.289999	0.414749	0.252145	0.366555	0.373425	0.257847	...	0.308020	-0.022624	-0.050077	0.336410	-0.306834	-0.640987	1.000000	-0.465793	0.011286	-0.011286
InternetService_No	-0.182742	0.000615	0.139812	-0.039062	0.172209	-0.082560	-0.797084	-0.803663	-0.803500	-0.797300	...	-0.227890	-0.002113	0.001030	-0.284917	0.321361	-0.380635	-0.465793	1.000000	-0.006026	0.006026
gender_Female	0.001874	0.001808	-0.010517	-0.005106	0.006488	0.009451	0.014418	0.012230	0.004720	0.009409	...	0.008612	0.016024	-0.001215	-0.000752	-0.013744	-0.006568	0.011286	-0.006026	1.000000	-1.000000
gender_Male	-0.001874	-0.001808	0.010517	0.005106	-0.006488	-0.009451	-0.014418	-0.012230	-0.004720	-0.009409	...	-0.008612	-0.016024	0.001215	0.000752	0.013744	0.006568	-0.011286	0.006026	-1.000000	1.000000

26 rows × 26 columns

sns.set(rc={'figure.figsize':(15.7,8)})

sns.set(style="ticks", color_codes=True)

sns.heatmap(new_df.corr(), annot=True, linewidths=0.5, center=0, cbar=False, cmap="YlGnBu")

<AxesSubplot:>

# 2F...... Splitting data into 80% Train and 20% Test

X = new_df.drop(["customerID","Churn"], axis=1)   #........... Independent variables, 
y = new_df['Churn']                           #.............. Dependent variable

X.dtypes

SeniorCitizen                                int64
Partner                                      int64
Dependents                                   int64
tenure                                       int64
PhoneService                                 int64
MultipleLines                                int64
OnlineSecurity                               int64
OnlineBackup                                 int64
DeviceProtection                             int64
TechSupport                                  int64
StreamingTV                                  int64
StreamingMovies                              int64
Contract                                     int64
PaperlessBilling                             int64
MonthlyCharges                             float64
TotalCharges                               float64
PaymentMethod_Bank transfer (automatic)      uint8
PaymentMethod_Credit card (automatic)        uint8
PaymentMethod_Electronic check               uint8
PaymentMethod_Mailed check                   uint8
InternetService_DSL                          uint8
InternetService_Fiber optic                  uint8
InternetService_No                           uint8
gender_Female                                uint8
gender_Male                                  uint8
dtype: object

X.describe()

	SeniorCitizen	Partner	Dependents	tenure	PhoneService	MultipleLines	OnlineSecurity	OnlineBackup	DeviceProtection	TechSupport	...	TotalCharges	PaymentMethod_Bank transfer (automatic)	PaymentMethod_Credit card (automatic)	PaymentMethod_Electronic check	PaymentMethod_Mailed check	InternetService_DSL	InternetService_Fiber optic	InternetService_No	gender_Female	gender_Male
count	7043.000000	7043.000000	7043.000000	7043.000000	7043.000000	7043.000000	7043.000000	7043.000000	7043.000000	7043.000000	...	7043.000000	7043.000000	7043.000000	7043.000000	7043.000000	7043.000000	7043.000000	7043.000000	7043.000000	7043.000000
mean	0.162147	0.483033	0.299588	32.371149	0.903166	0.325004	0.069999	0.128212	0.127219	0.073548	...	2281.916928	0.219225	0.216101	0.335794	0.228880	0.343746	0.439585	0.216669	0.495244	0.504756
std	0.368612	0.499748	0.458110	24.559481	0.295752	0.642730	0.706051	0.738369	0.737868	0.708201	...	2265.270398	0.413751	0.411613	0.472301	0.420141	0.474991	0.496372	0.412004	0.500013	0.500013
min	0.000000	0.000000	0.000000	0.000000	0.000000	-1.000000	-1.000000	-1.000000	-1.000000	-1.000000	...	18.800000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000
25%	0.000000	0.000000	0.000000	9.000000	1.000000	0.000000	0.000000	0.000000	0.000000	0.000000	...	402.225000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000
50%	0.000000	0.000000	0.000000	29.000000	1.000000	0.000000	0.000000	0.000000	0.000000	0.000000	...	1397.475000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	1.000000
75%	0.000000	1.000000	1.000000	55.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	...	3786.600000	0.000000	0.000000	1.000000	0.000000	1.000000	1.000000	0.000000	1.000000	1.000000
max	1.000000	1.000000	1.000000	72.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	...	8684.800000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000

8 rows × 25 columns

from sklearn import preprocessing

X_standard = preprocessing.scale(X)
X_standard

array([[-0.43991649,  1.03453023, -0.65401193, ..., -0.52592737,
         1.00955867, -1.00955867],
       [-0.43991649, -0.96662231, -0.65401193, ..., -0.52592737,
        -0.99053183,  0.99053183],
       [-0.43991649, -0.96662231, -0.65401193, ..., -0.52592737,
        -0.99053183,  0.99053183],
       ...,
       [-0.43991649,  1.03453023,  1.5290241 , ..., -0.52592737,
         1.00955867, -1.00955867],
       [ 2.27315869,  1.03453023, -0.65401193, ..., -0.52592737,
        -0.99053183,  0.99053183],
       [-0.43991649, -0.96662231, -0.65401193, ..., -0.52592737,
        -0.99053183,  0.99053183]])

from sklearn.model_selection import train_test_split

X_train, X_test, y_train, y_test = train_test_split(X_standard, y, test_size=.20, random_state=1)

print("X_train shape",X_train.shape)
print("y_train shape",y_train.shape)
print("X_test shape",X_test.shape)
print("y_test shape",y_test.shape)

X_train shape (5634, 25)
y_train shape (5634,)
X_test shape (1409, 25)
y_test shape (1409,)

X_train

array([[-0.43991649,  1.03453023,  1.5290241 , ...,  1.90140322,
        -0.99053183,  0.99053183],
       [-0.43991649, -0.96662231, -0.65401193, ..., -0.52592737,
         1.00955867, -1.00955867],
       [-0.43991649,  1.03453023, -0.65401193, ..., -0.52592737,
        -0.99053183,  0.99053183],
       ...,
       [-0.43991649, -0.96662231, -0.65401193, ..., -0.52592737,
        -0.99053183,  0.99053183],
       [-0.43991649, -0.96662231, -0.65401193, ..., -0.52592737,
        -0.99053183,  0.99053183],
       [-0.43991649,  1.03453023,  1.5290241 , ..., -0.52592737,
         1.00955867, -1.00955867]])

#....Tree-based models care more about the order of values than about the absolute value that a feature takes. 
#....Hence, normalization are skipped for decision trees and used mainly in linear models/KNN/neural networks because 
#....they’re affected by absolute values taken by features.

#.......Decision trees are also not sensitive to outliers since the partitioning happens based on the proportion of samples 
#.......within the split ranges and not on absolute values.

#.......Hence removing the scaling factor and going back to regulary X

X = new_df.drop(["customerID","Churn"], axis=1)   #........... Independent variables, 
y = new_df['Churn']                           #.............. Dependent variable

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=.20, random_state=1)

X_train

	SeniorCitizen	Partner	Dependents	tenure	PhoneService	MultipleLines	OnlineSecurity	OnlineBackup	DeviceProtection	TechSupport	...	TotalCharges	PaymentMethod_Bank transfer (automatic)	PaymentMethod_Credit card (automatic)	PaymentMethod_Electronic check	PaymentMethod_Mailed check	InternetService_DSL	InternetService_Fiber optic	InternetService_No	gender_Female	gender_Male
1814	0	1	1	12	1	0	-1	-1	-1	-1	...	258.35	0	0	0	1	0	0	1	0	1
5946	0	0	0	42	1	0	1	1	1	1	...	3160.55	0	1	0	0	1	0	0	1	0
3881	0	1	0	71	1	1	1	1	0	1	...	4681.75	1	0	0	0	1	0	0	0	1
2389	0	1	1	71	1	1	1	0	1	1	...	6300.85	0	0	1	0	1	0	0	0	1
3676	0	0	0	30	1	0	1	1	0	1	...	2044.75	0	0	1	0	1	0	0	0	1
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
905	1	0	0	9	1	1	0	0	1	0	...	918.60	0	0	1	0	0	1	0	0	1
5192	0	0	1	60	1	0	-1	-1	-1	-1	...	1189.90	0	0	0	1	0	0	1	0	1
3980	0	0	0	28	1	1	0	1	1	0	...	2979.50	0	0	1	0	0	1	0	0	1
235	0	0	0	2	1	1	0	1	0	0	...	114.10	0	0	0	1	1	0	0	0	1
5157	0	1	1	16	1	0	1	1	0	1	...	1114.85	1	0	0	0	1	0	0	1	0

5634 rows × 25 columns

y_train

1814    0
5946    1
3881    0
2389    0
3676    0
       ..
905     1
5192    0
3980    1
235     1
5157    0
Name: Churn, Length: 5634, dtype: int64

from sklearn.tree import DecisionTreeClassifier

dt_model = DecisionTreeClassifier(criterion = 'gini', random_state=1)
dt_model.fit(X_train, y_train)

DecisionTreeClassifier(random_state=1)

print(dt_model.score(X_train, y_train))
print(dt_model.score(X_test, y_test))

0.9980475683351083
0.7388218594748048

y_pred_0 = dt_model.predict(X_test)

from sklearn import metrics
from sklearn.metrics import roc_auc_score

cm=metrics.confusion_matrix(y_test, y_pred_0, labels=[1, 0])

df_cm = pd.DataFrame(cm, index = [i for i in ["1","0"]],
                  columns = [i for i in ["Predict 1","Predict 0"]])
plt.figure(figsize = (9,7))
sns.heatmap(df_cm, annot=True)

<AxesSubplot:>

print("Decision Tree - Accuracy ",metrics.accuracy_score(y_test, y_pred_0))
print("Decision Tree - Precision",metrics.precision_score(y_test, y_pred_0))
print("Decision Tree - Recall   ",metrics.recall_score(y_test, y_pred_0))
print("Decision Tree - F1 score ",metrics.f1_score(y_test, y_pred_0))
print("Decision Tree - ROC      ",metrics.roc_auc_score(y_test, y_pred_0))

Decision Tree - Accuracy  0.7388218594748048
Decision Tree - Precision 0.47512437810945274
Decision Tree - Recall    0.5488505747126436
Decision Tree - F1 score  0.5093333333333334
Decision Tree - ROC       0.6749907915976036

X_train

	SeniorCitizen	Partner	Dependents	tenure	PhoneService	MultipleLines	OnlineSecurity	OnlineBackup	DeviceProtection	TechSupport	...	TotalCharges	PaymentMethod_Bank transfer (automatic)	PaymentMethod_Credit card (automatic)	PaymentMethod_Electronic check	PaymentMethod_Mailed check	InternetService_DSL	InternetService_Fiber optic	InternetService_No	gender_Female	gender_Male
1814	0	1	1	12	1	0	-1	-1	-1	-1	...	258.35	0	0	0	1	0	0	1	0	1
5946	0	0	0	42	1	0	1	1	1	1	...	3160.55	0	1	0	0	1	0	0	1	0
3881	0	1	0	71	1	1	1	1	0	1	...	4681.75	1	0	0	0	1	0	0	0	1
2389	0	1	1	71	1	1	1	0	1	1	...	6300.85	0	0	1	0	1	0	0	0	1
3676	0	0	0	30	1	0	1	1	0	1	...	2044.75	0	0	1	0	1	0	0	0	1
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
905	1	0	0	9	1	1	0	0	1	0	...	918.60	0	0	1	0	0	1	0	0	1
5192	0	0	1	60	1	0	-1	-1	-1	-1	...	1189.90	0	0	0	1	0	0	1	0	1
3980	0	0	0	28	1	1	0	1	1	0	...	2979.50	0	0	1	0	0	1	0	0	1
235	0	0	0	2	1	1	0	1	0	0	...	114.10	0	0	0	1	1	0	0	0	1
5157	0	1	1	16	1	0	1	1	0	1	...	1114.85	1	0	0	0	1	0	0	1	0

5634 rows × 25 columns

from sklearn.tree import export_graphviz

train_char_label = ['No', 'Yes']
Credit_Tree_File = open('credit_tree.dot','w')
dot_data = export_graphviz(dt_model, out_file=Credit_Tree_File, feature_names = list(X_train), class_names = list(train_char_label))
Credit_Tree_File.close()

from os import system
from IPython.display import Image

#Works only if "dot" command works on you machine

retCode = system("dot -Tpng credit_tree.dot -o credit_tree.png")
if(retCode>0):
    print("system command returning error: "+str(retCode))
else:
    display(Image("credit_tree.png"))

3. Model building and Improvement

!pip install xgboost

Requirement already satisfied: xgboost in c:\users\hp\anaconda3\lib\site-packages (1.5.2)
Requirement already satisfied: scipy in c:\users\hp\anaconda3\lib\site-packages (from xgboost) (1.6.2)
Requirement already satisfied: numpy in c:\users\hp\anaconda3\lib\site-packages (from xgboost) (1.20.1)

# 3A.......... Train model using XGBoost

import xgboost as xgb
from xgboost.sklearn import XGBClassifier

#from sklearn.grid_search import GridSearchCV   #Perforing grid search

xgb_mod = XGBClassifier(use_label_encoder=False)

xgb_mod.fit(X_train,y_train)

[10:44:44] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.

XGBClassifier(base_score=0.5, booster='gbtree', colsample_bylevel=1,
              colsample_bynode=1, colsample_bytree=1, enable_categorical=False,
              gamma=0, gpu_id=-1, importance_type=None,
              interaction_constraints='', learning_rate=0.300000012,
              max_delta_step=0, max_depth=6, min_child_weight=1, missing=nan,
              monotone_constraints='()', n_estimators=100, n_jobs=8,
              num_parallel_tree=1, predictor='auto', random_state=0,
              reg_alpha=0, reg_lambda=1, scale_pos_weight=1, subsample=1,
              tree_method='exact', use_label_encoder=False,
              validate_parameters=1, verbosity=None)

print(xgb_mod.score(X_train,y_train))
print(xgb_mod.score(X_test,y_test))

0.9366347177848775
0.8019872249822569

y_pred_xgb = xgb_mod.predict(X_test)

from sklearn import metrics
from sklearn.metrics import roc_auc_score

cm=metrics.confusion_matrix(y_test, y_pred_xgb, labels=[1, 0])

df_cm = pd.DataFrame(cm, index = [i for i in ["1","0"]],
                  columns = [i for i in ["Predict 1","Predict 0"]])
plt.figure(figsize = (9,7))
sns.heatmap(df_cm, annot=True)

print("with XGBoost - Accuracy ",metrics.accuracy_score(y_test, y_pred_xgb))
print("with XGBoost - Precision",metrics.precision_score(y_test, y_pred_xgb))
print("with XGBoost - Recall   ",metrics.recall_score(y_test, y_pred_xgb))
print("with XGBoost - F1 score ",metrics.f1_score(y_test, y_pred_xgb))
print("with XGBoost - ROC      ",metrics.roc_auc_score(y_test, y_pred_xgb))

perf_cmp_Df = pd.DataFrame()

temp_perf_Df = pd.DataFrame({'XGBoost Classifier':["Base model"],'Accuracy': [metrics.accuracy_score(y_test, y_pred_xgb)],'Precision': [metrics.precision_score(y_test, y_pred_xgb)],'Recall': [metrics.recall_score(y_test, y_pred_xgb)],'F1 score': [metrics.f1_score(y_test, y_pred_xgb)],'ROC score': [metrics.roc_auc_score(y_test, y_pred_xgb)]})
perf_cmp_Df = pd.concat([perf_cmp_Df, temp_perf_Df])
perf_cmp_Df = perf_cmp_Df[['XGBoost Classifier','Accuracy','Precision','Recall','F1 score','ROC score']]
perf_cmp_Df            

with XGBoost - Accuracy  0.8019872249822569
with XGBoost - Precision 0.6029850746268657
with XGBoost - Recall    0.5804597701149425
with XGBoost - F1 score  0.5915080527086384
with XGBoost - ROC       0.7275531649820707

	XGBoost Classifier	Accuracy	Precision	Recall	F1 score	ROC score
0	Base model	0.801987	0.602985	0.58046	0.591508	0.727553

# 3B....... Improve performance of XGBoost 

xgb_mod = XGBClassifier(
 learning_rate =0.1,
 n_estimators=200,
 max_depth=5,
 min_child_weight=1,
 gamma=0,
 subsample=0.8,
 colsample_bytree=0.8,
 objective='binary:logistic',
 nthread=4,
 use_label_encoder=False,
 scale_pos_weight=1,seed=27)

xgb_mod.fit(X_train, y_train)

print(xgb_mod.score(X_train,y_train))
print(xgb_mod.score(X_test,y_test))

y_pred_xgb = xgb_mod.predict(X_test)

cm=metrics.confusion_matrix(y_test, y_pred_xgb, labels=[1, 0])

df_cm = pd.DataFrame(cm, index = [i for i in ["1","0"]],
                  columns = [i for i in ["Predict 1","Predict 0"]])
plt.figure(figsize = (9,7))
sns.heatmap(df_cm, annot=True)

print("Tuning 1 XGBoost - Accuracy ",metrics.accuracy_score(y_test, y_pred_xgb))
print("Tuning 1 XGBoost - Precision",metrics.precision_score(y_test, y_pred_xgb))
print("Tuning 1 XGBoost - Recall   ",metrics.recall_score(y_test, y_pred_xgb))
print("Tuning 1 XGBoost - F1 score ",metrics.f1_score(y_test, y_pred_xgb))
print("Tuning 1 XGBoost - ROC      ",metrics.roc_auc_score(y_test, y_pred_xgb))

temp_perf_Df = pd.DataFrame({'XGBoost Classifier':["l_rate =0.1,esti=200,depth=5,ch_weight=1,gamma=0,s_bytree=0.8"],
                             'Accuracy': [metrics.accuracy_score(y_test, y_pred_xgb)],
                             'Precision': [metrics.precision_score(y_test, y_pred_xgb)],
                             'Recall': [metrics.recall_score(y_test, y_pred_xgb)],
                             'F1 score': [metrics.f1_score(y_test, y_pred_xgb)],
                             'ROC score': [metrics.roc_auc_score(y_test, y_pred_xgb)]})
perf_cmp_Df = pd.concat([perf_cmp_Df, temp_perf_Df])
perf_cmp_Df = perf_cmp_Df[['XGBoost Classifier','Accuracy','Precision','Recall','F1 score','ROC score']]
perf_cmp_Df            

[10:45:23] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.
0.88604898828541
0.8076650106458482
Tuning 1 XGBoost - Accuracy  0.8076650106458482
Tuning 1 XGBoost - Precision 0.6199376947040498
Tuning 1 XGBoost - Recall    0.5718390804597702
Tuning 1 XGBoost - F1 score  0.5949177877428998
Tuning 1 XGBoost - ROC       0.7284266090329011

	XGBoost Classifier	Accuracy	Precision	Recall	F1 score	ROC score
0	Base model	0.801987	0.602985	0.580460	0.591508	0.727553
0	l_rate =0.1,esti=200,depth=5,ch_weight=1,gamma...	0.807665	0.619938	0.571839	0.594918	0.728427

from sklearn.model_selection import cross_validate
from sklearn.model_selection import GridSearchCV
from sklearn.model_selection import StratifiedKFold

xgb_model = XGBClassifier(use_label_encoder=False,eval_metric="logloss")

#List Hyperparameters to tune

eta = [0.01, 0.05, 0.10, 0.15]
gamma = [0, 0.1, 0.2, 0.3]
max_depth = [3, 4, 5]
min_ch_wgt= [1,3,5]
col_sam_tree= [0.3,0.4,0.5]

#convert to dictionary
hyperparameters = dict(learning_rate=eta, gamma=gamma, max_depth=max_depth, min_child_weight = min_ch_wgt, colsample_bytree=col_sam_tree)
#Making model
clf = GridSearchCV(xgb_model, hyperparameters, cv=5)
best_model = clf.fit(X_train,y_train)

#Best Hyperparameters Value
print("")
print('Best learning rate:', best_model.best_estimator_.get_params()['learning_rate'])
print('Best gamma:', best_model.best_estimator_.get_params()['gamma'])
print('Best max_depth:', best_model.best_estimator_.get_params()['max_depth'])
print('Best min_child_weight:', best_model.best_estimator_.get_params()['min_child_weight'])
print('Best colsample_bytree:', best_model.best_estimator_.get_params()['colsample_bytree'])


print("")
#Predict testing set
y_pred = best_model.predict(X_test)

print("GridSearchCV - Accuracy ",metrics.accuracy_score(y_test, y_pred))
print("GridSearchCV - Precision",metrics.precision_score(y_test, y_pred, average='weighted'))
print("GridSearchCV - Recall   ",metrics.recall_score(y_test, y_pred, average='weighted'))
print("GridSearchCV - F1 score ",metrics.f1_score(y_test, y_pred, average='weighted'))

Best learning rate: 0.1
Best gamma: 0.1
Best max_depth: 3
Best min_child_weight: 5
Best colsample_bytree: 0.5

GridSearchCV - Accuracy  0.8133427963094393
GridSearchCV - Precision 0.8094690029446563
GridSearchCV - Recall    0.8133427963094393
GridSearchCV - F1 score  0.811153986406265

# 3B....... Best performing hyperparameters are

#          Best learning rate: 0.1
#          Best gamma: 0.1
#          Best max_depth: 3
#          Best min_child_weight: 5
#          Best colsample_bytree: 0.5
    
#          and the best test performance is

#          GridSearchCV - Accuracy  0.8133427963094393
#          GridSearchCV - Precision 0.8094690029446563
#          GridSearchCV - Recall    0.8133427963094393
#          GridSearchCV - F1 score  0.811153986406265  

xgb = XGBClassifier(
 learning_rate =0.1,
 n_estimators=450,
 max_depth=3,
 min_child_weight=5,
 gamma=0.1,
 subsample=0.8,
 colsample_bytree=0.5,
 objective='binary:logistic',
 nthread=4,
 use_label_encoder=False,
 eval_metric="logloss",
 scale_pos_weight=1,seed=27)

xgb.fit(X_train, y_train)

print(xgb.score(X_train,y_train))
print(xgb.score(X_test,y_test))

y_pred_xgb = xgb.predict(X_test)

cm=metrics.confusion_matrix(y_test, y_pred_xgb, labels=[1, 0])

df_cm = pd.DataFrame(cm, index = [i for i in ["1","0"]],
                  columns = [i for i in ["Predict 1","Predict 0"]])
plt.figure(figsize = (9,7))
sns.heatmap(df_cm, annot=True)

print("Tuning 2 XGBoost - Accuracy ",metrics.accuracy_score(y_test, y_pred_xgb))
print("Tuning 2 XGBoost - Precision",metrics.precision_score(y_test, y_pred_xgb))
print("Tuning 2 XGBoost - Recall   ",metrics.recall_score(y_test, y_pred_xgb))
print("Tuning 2 XGBoost - F1 score ",metrics.f1_score(y_test, y_pred_xgb))
print("Tuning 2 XGBoost - ROC      ",metrics.roc_auc_score(y_test, y_pred_xgb))

temp_perf_Df = pd.DataFrame({'XGBoost Classifier':["l_rate =0.0.5,esti=250,depth=3,ch_weight=3,gamma=0.2,s_bytree=0.4"],
                             'Accuracy': [metrics.accuracy_score(y_test, y_pred_xgb)],
                             'Precision': [metrics.precision_score(y_test, y_pred_xgb)],
                             'Recall': [metrics.recall_score(y_test, y_pred_xgb)],
                             'F1 score': [metrics.f1_score(y_test, y_pred_xgb)],
                             'ROC score': [metrics.roc_auc_score(y_test, y_pred_xgb)]})
perf_cmp_Df = pd.concat([perf_cmp_Df, temp_perf_Df])
perf_cmp_Df = perf_cmp_Df[['XGBoost Classifier','Accuracy','Precision','Recall','F1 score','ROC score']]
perf_cmp_Df            

0.8462903798367057
0.8041163946061036
Tuning 2 XGBoost - Accuracy  0.8041163946061036
Tuning 2 XGBoost - Precision 0.6104294478527608
Tuning 2 XGBoost - Recall    0.5718390804597702
Tuning 2 XGBoost - F1 score  0.5905044510385757
Tuning 2 XGBoost - ROC       0.7260703413608935

	XGBoost Classifier	Accuracy	Precision	Recall	F1 score	ROC score
0	Base model	0.801987	0.602985	0.580460	0.591508	0.727553
0	l_rate =0.1,esti=200,depth=5,ch_weight=1,gamma...	0.807665	0.619938	0.571839	0.594918	0.728427
0	l_rate =0.0.5,esti=250,depth=3,ch_weight=3,gam...	0.805536	0.612805	0.577586	0.594675	0.728944
0	l_rate =0.0.5,esti=250,depth=3,ch_weight=3,gam...	0.804116	0.610429	0.571839	0.590504	0.726070

from sklearn.metrics import mean_squared_error

xg_reg = xgb.XGBRegressor(objective ='reg:linear', colsample_bytree = 0.3, learning_rate = 0.1,
                max_depth = 5, alpha = 10, n_estimators = 100)
xg_reg.fit(X_train,y_train)

preds = xg_reg.predict(X_test)

rmse = np.sqrt(mean_squared_error(y_test, preds))
print("RMSE: %f" % (rmse))

[14:43:29] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/objective/regression_obj.cu:188: reg:linear is now deprecated in favor of reg:squarederror.
RMSE: 0.354577

data_dmatrix = xgb.DMatrix(data=X,label=y)

params = {"objective":"reg:linear",'colsample_bytree': 0.3,'learning_rate': 0.1,
                'max_depth': 5, 'alpha': 10}

cv_results = xgb.cv(dtrain=data_dmatrix, params=params, nfold=3,
                    num_boost_round=50,early_stopping_rounds=10,metrics="rmse", as_pandas=True, seed=123)

[14:43:32] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/objective/regression_obj.cu:188: reg:linear is now deprecated in favor of reg:squarederror.
[14:43:32] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/objective/regression_obj.cu:188: reg:linear is now deprecated in favor of reg:squarederror.
[14:43:32] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/objective/regression_obj.cu:188: reg:linear is now deprecated in favor of reg:squarederror.

cv_results.head(10)

	train-rmse-mean	train-rmse-std	test-rmse-mean	test-rmse-std
0	0.479932	0.000444	0.480428	0.000266
1	0.464723	0.001017	0.465395	0.001602
2	0.452284	0.001441	0.453515	0.000340
3	0.439544	0.001557	0.441052	0.001102
4	0.429754	0.002944	0.431591	0.002260
5	0.421163	0.001664	0.423180	0.002409
6	0.412995	0.001716	0.415241	0.002067
7	0.405852	0.001942	0.408474	0.002092
8	0.400392	0.001260	0.403382	0.002611
9	0.395491	0.002103	0.398697	0.002633

print((cv_results["test-rmse-mean"]).tail(1))

49    0.368023
Name: test-rmse-mean, dtype: float64

xg_reg = xgb.train(params=params, dtrain=data_dmatrix, num_boost_round=10)

[14:43:46] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/objective/regression_obj.cu:188: reg:linear is now deprecated in favor of reg:squarederror.

from sklearn.ensemble import GradientBoostingClassifier

gbcl = GradientBoostingClassifier(n_estimators = 100, learning_rate = 0.05, random_state=1)
gbcl = gbcl.fit(X_train, y_train)

print(gbcl.score(X_train,y_train))
print(gbcl.score(X_test,y_test))

y_pred_4 = gbcl.predict(X_test)

cm=metrics.confusion_matrix(y_test, y_pred_4, labels=[1, 0])

df_cm = pd.DataFrame(cm, index = [i for i in ["1","0"]],
                  columns = [i for i in ["Predict 1","Predict 0"]])
plt.figure(figsize = (9,7))
sns.heatmap(df_cm, annot=True)

print("Trees - 100, rate=0.05 XGB - Accuracy ",metrics.accuracy_score(y_test, y_pred_4))
print("Trees - 100, rate=0.05 XGB - Precision",metrics.precision_score(y_test, y_pred_4))
print("Trees - 100, rate=0.05 XGB - Recall   ",metrics.recall_score(y_test, y_pred_4))
print("Trees - 100, rate=0.05 XGB - F1 score ",metrics.f1_score(y_test, y_pred_4))
print("Trees - 100, rate=0.05 XGB - ROC      ",metrics.roc_auc_score(y_test, y_pred_4))

0.8143414980475683
0.8034066713981547
Trees - 100, rate=0.05 XGB - Accuracy  0.8034066713981547
Trees - 100, rate=0.05 XGB - Precision 0.6187290969899666
Trees - 100, rate=0.05 XGB - Recall    0.5316091954022989
Trees - 100, rate=0.05 XGB - F1 score  0.5718701700154559
Trees - 100, rate=0.05 XGB - ROC       0.7120816947793775

gbcl = GradientBoostingClassifier(n_estimators = 125, learning_rate = 0.08, random_state=1)
gbcl = gbcl.fit(X_train, y_train)

print(gbcl.score(X_train,y_train))
print(gbcl.score(X_test,y_test))

y_pred_5 = gbcl.predict(X_test)

cm=metrics.confusion_matrix(y_test, y_pred_5, labels=[1, 0])

df_cm = pd.DataFrame(cm, index = [i for i in ["1","0"]],
                  columns = [i for i in ["Predict 1","Predict 0"]])
plt.figure(figsize = (9,7))
sns.heatmap(df_cm, annot=True)

print("Trees - 125, rate=0.08 XGB - Accuracy ",metrics.accuracy_score(y_test, y_pred_5))
print("Trees - 125, rate=0.08 XGB - Precision",metrics.precision_score(y_test, y_pred_5))
print("Trees - 125, rate=0.08 XGB - Recall   ",metrics.recall_score(y_test, y_pred_5))
print("Trees - 125, rate=0.08 XGB - F1 score ",metrics.f1_score(y_test, y_pred_5))
print("Trees - 125, rate=0.08 XGB - ROC      ",metrics.roc_auc_score(y_test, y_pred_5))

0.8230386936457224
0.808374733853797
Trees - 125, rate=0.08 XGB - Accuracy  0.808374733853797
Trees - 125, rate=0.08 XGB - Precision 0.6242038216560509
Trees - 125, rate=0.08 XGB - Recall    0.5632183908045977
Trees - 125, rate=0.08 XGB - F1 score  0.5921450151057401
Trees - 125, rate=0.08 XGB - ROC       0.7260012783429208

Part B - Domain IT

#  working only with the TelecomCustomers-Churn_2.csv as a part of modularisation.

    # importing all the necessary libraries

    import numpy as np
    import pandas as pd
    import matplotlib.pyplot as plt
    %matplotlib inline
    import seaborn as sns
    
    from sklearn import preprocessing
    from sklearn.model_selection import train_test_split
    
    #.. importing the classifier libraries    
    
    from sklearn.tree import DecisionTreeClassifier
    from sklearn.ensemble import GradientBoostingClassifier
    import xgboost as xgb
    from xgboost.sklearn import XGBClassifier
    from sklearn.neighbors import KNeighborsClassifier
    from sklearn.linear_model import LogisticRegression
    from sklearn.svm import SVC  

    # importing the metrics libraries
    
    from sklearn import metrics
    from sklearn.metrics import roc_auc_score
    from sklearn.metrics import mean_squared_error

    # importing the tree libraries
    
    from sklearn.tree import export_graphviz
    from os import system
    from IPython.display import Image
    
    import pickle
    
    global np,pd,plt,sns, xgb
    
    global resultsDf
            
    def csv(file):  #  called from data_under function
        
        print("passed file",file) 
        global data
    
        data = pd.read_csv(file) # loading the csv file into a dataframe
        data.head(10)
        
        return
    
    def details(df):  # called from data_under function
        
        print("Datatypes :",df.dtypes)  # printing the datatypes
        print("")
        print("Dataframe shape :",df.shape)  # printing the shape ------ row, columns
        print("")
        print("Describe",df.describe())   # printing the 5 point summary
        print("")
        print("sample records",df.head(10))
    
    def data_under(file):
        
        csv(file)  # passing a csv file as an argument to read into a dataframe
        details(data) # printing the datatypes, shape and 5 point summary
    
    def null_check(data):  # to check if there are any missing values / nulls and fill them up with median values.
        
        data.isnull().any().sum()
        data.TotalCharges = pd.to_numeric(data.TotalCharges, errors="coerce")
        data.isnull().values.any()
        median = data['TotalCharges'].median()  # this column has 11 missing values and is updated with median values
        data['TotalCharges'].fillna(median, inplace=True)  # dataframe filled with median values
    
        return data
    
    def show_data(data):  # to check the distribution of categorical column data

        for i in data.columns:
            if ((i in ["customerID","MonthlyCharges","TotalCharges"])):  # since they are not categorical columns
                print()
            else:   
                print("% S unique values", i)
                print(data[i].value_counts())
                print("")
    
    def prep_exp(data):
        
        null_check(data)
        show_data(data)
    
    def hist(data): # to print histograms for all the categorical columns

        for i in data.columns:
            if ((i in ["customerID","MonthlyCharges","TotalCharges"])):  # not considering since they are not categorical
                print()
            else:   
                fig = plt.figure(figsize = (10,8))
                ax = fig.gca()
                plt.figure(i)
                sns.histplot(data[i], kde=True, ax=ax)
                            
    def pie_new(data): # printing pie charts for all the categorical columns
        
        for i in data.columns:
            if (data[i].dtypes == "object"):
            
                if i in ["customerID","MonthlyCharges","TotalCharges"]:
                    print()
                else:
                    print("column is %S",i)                                
                    data.groupby(i).size().plot(kind='pie', subplots=True, shadow=True, startangle=30, figsize=(8,6), autopct='%1.2f%%')                
                    plt.tight_layout()
                    plt.show()
    
    def encode(data): # to encode the categorical variables
        
        global new_data
        replaceSt = {
        "OnlineBackup":      {"Yes": 1, "No": 0 ,"No internet service": -1},
        "DeviceProtection":  {"Yes": 1, "No": 0 ,"No internet service": -1},
        "TechSupport":       {"Yes": 1, "No": 0 ,"No internet service": -1},
        "StreamingTV":       {"Yes": 1, "No": 0 ,"No internet service": -1},
        "StreamingMovies":   {"Yes": 1, "No": 0 ,"No internet service": -1},
        "Contract":          {"Month-to-month": 1, "One year": 2 ,"Two year": 3},
        "Churn":             {"Yes": 1, "No": 0 },                
        "PaperlessBilling":  {"Yes": 1, "No": 0 } 
        }
        oneHot=["PaymentMethod"] # since ordering payment methods isnt possible, oneHot encoding it - creating multiple columns 
    
        new_data=data.replace(replaceSt)
        new_data=pd.get_dummies(new_data, columns=oneHot)  # new_data dataframe with encoded data
        new_data.head(10)
    
        return(new_data)

    def cor(new_data): # function to create correlation among all the variables and printing heatmap

        sns.set(rc={'figure.figsize':(15.7,8)})
        sns.set(style="ticks", color_codes=True)
        font1 = {'family':'serif','color':'blue','size':20}
        plt.title("Corelation", fontdict = font1)
        sns.heatmap(new_data.corr(), annot=True, linewidths=0.5, center=0, cbar=False, cmap="YlGnBu")
        
    def analysis(data):
        hist(data)
        pie_new(data)
        encode(data)
        cor(new_data)
    
    def split(new_data):  # splitting the Independent and Dependent variables 
    
        global X,y,X_train1,y_train1,X_test1,y_test1
        
        X = new_data.drop(["customerID","Churn"], axis=1)   #........... Independent variables, 
        y = new_data['Churn']                               #.............. Dependent variable
                    
        X_train1, X_test1, y_train1, y_test1 = train_test_split(X, y, test_size=.20, random_state=2)
               
        return
    
    def export_tree(dt_model):  # visualizing a tree picture for the decision tree

        train_char_label = ['No', 'Yes']
        Credit_Tree_File = open('IT_credit_tree.dot','w')
        dot_data = export_graphviz(dt_model, out_file=Credit_Tree_File, feature_names = list(X_train1), class_names = list(train_char_label))
        Credit_Tree_File.close()
    
        retCode = system("dot -Tpng IT_credit_tree.dot -o IT_credit_tree.png")
        if(retCode>0):
            print("system command returning error: "+str(retCode))
        else:
            print("Displaying tree image")
            display(Image("IT_credit_tree.png"))   # displaying the tree image
    
    def un_pikling(X_test1,y_test1):   # unpickling the GradientBoosting model and predicting the dependent variable
    
        # load model
        with open('model.pkl', 'rb') as f:
            loaded_classifier = pickle.load(f)

        # predict
        y_true = y_test1
        y_pred = loaded_classifier.predict(X_test1)
        print("")
        print('pickle for Gradientclassifier: accuracy:', metrics.accuracy_score(y_true, y_pred))
    
    
    def mod_fit(X_train1,y_train1,X_test1,y_test1):  # Calling multiple classifier models to fit the training data
        
        mod = [DecisionTreeClassifier(criterion = 'gini', random_state=2),
       GradientBoostingClassifier(n_estimators = 75, learning_rate = 0.04, random_state=1),
       XGBClassifier(learning_rate =0.1,n_estimators=200,max_depth=5,min_child_weight=1,gamma=0,subsample=0.8,colsample_bytree=0.8,objective='binary:logistic',nthread=4,scale_pos_weight=1,eed=27),
       KNeighborsClassifier(n_neighbors= 5, weights = 'distance' ),
       LogisticRegression(solver='liblinear'),
       SVC(kernel='linear')]

        global resultsDf
        
        resultsDf = pd.DataFrame()
        
        for i in mod:
            
            model = i
            
            new_i = str(i)  # Classification models passed need to be converetd to string
            sub_i = new_i[0:new_i.find("(")]   # to pick the leftmost characters before the "(" paranthesis
            
            if sub_i == "SVC":   # standardization being carried out only for SVC model
                
                X_stan = preprocessing.scale(X)
                X_train1, X_test1, y_train1, y_test1 = train_test_split(X_stan, y, test_size=.20, random_state=2)
                
            elif sub_i == "DecisionTreeClassifier":     # Tree visualization created only for Decision tree classifier          
                
                model.fit(X_train1, y_train1)
                export_tree(model)
                
            elif sub_i == "GradientBoostingClassifier":  
                
                # Since this is the best performing model creating a pickle file for it
                
                model.fit(X_train1, y_train1)
                pickle.dump(model, open('model.pkl','wb'))  # writing the model to a pickle file in binary mode.
                
                                        
            model.fit(X_train1, y_train1)
            
            print("")
    
            y_pred = model.predict(X_test1)
    
            cm=metrics.confusion_matrix(y_test1, y_pred, labels=[1, 0])   # Creatng a confusion matrix to check the metrics

            df_cm = pd.DataFrame(cm, index = [i for i in ["1","0"]],
                  columns = [i for i in ["Predict 1","Predict 0"]])
            plt.figure(figsize = (9,7))
            font1 = {'family':'serif','color':'blue','size':20}
            plt.title(sub_i, fontdict = font1)
            sns.heatmap(df_cm, annot=True, linewidths=0.5, center=0, cbar=False, cmap="YlGnBu") 
            
            # collecting the metrics after each classifier prediction.
                                             
            tempResultsDf = pd.DataFrame({'Method':[sub_i],'Accuracy': [metrics.accuracy_score(y_test1, y_pred)],'Precision': [metrics.precision_score(y_test1, y_pred)],'Recall': [metrics.recall_score(y_test1, y_pred)],'F1 score': [metrics.f1_score(y_test1, y_pred)],'ROC score': [metrics.roc_auc_score(y_test1, y_pred)]})
            
            resultsDf = pd.concat([resultsDf, tempResultsDf])
            
            resultsDf = resultsDf[['Method','Accuracy','Precision','Recall','F1 score','ROC score']]
            
            
        return(resultsDf)
        
    def final_perf(resultDf):  # Checking the best performing classifier and printing metric graphs
        print("")
        print("Max accuracy")
        print(resultsDf['Accuracy'].max())
        
        print( resultsDf[resultsDf.loc[:,"Accuracy"] == resultsDf["Accuracy"].max() ] )
        
        resultsDf.plot.barh(x='Method', y='Accuracy', title='Accuracy comparison', color='orange')
        
        ax = resultsDf.plot.bar(rot=0)
                    
    def model_building(data):
    
        split(new_data)
        mod_fit(X_train1,y_train1,X_test1,y_test1)
                        
    def main(file):   # Main function which calls all the sub functions 
        
        data_under(file)
        prep_exp(data)
        analysis(data)
        model_building(data)
        details(resultsDf)
        final_perf(resultsDf)
        un_pikling(X_test1,y_test1)
        
        
        return
        
        if __name__ == "__main__":
            sys.exit(main())

main("TelcomCustomer-Churn_2.csv")

passed file TelcomCustomer-Churn_2.csv
Datatypes : customerID           object
OnlineBackup         object
DeviceProtection     object
TechSupport          object
StreamingTV          object
StreamingMovies      object
Contract             object
PaperlessBilling     object
PaymentMethod        object
MonthlyCharges      float64
TotalCharges         object
Churn                object
dtype: object

Dataframe shape : (7043, 12)

Describe        MonthlyCharges
count     7043.000000
mean        64.761692
std         30.090047
min         18.250000
25%         35.500000
50%         70.350000
75%         89.850000
max        118.750000

sample records    customerID OnlineBackup DeviceProtection TechSupport StreamingTV  \
0  7590-VHVEG          Yes               No          No          No   
1  5575-GNVDE           No              Yes          No          No   
2  3668-QPYBK          Yes               No          No          No   
3  7795-CFOCW           No              Yes         Yes          No   
4  9237-HQITU           No               No          No          No   
5  9305-CDSKC           No              Yes          No         Yes   
6  1452-KIOVK          Yes               No          No         Yes   
7  6713-OKOMC           No               No          No          No   
8  7892-POOKP           No              Yes         Yes         Yes   
9  6388-TABGU          Yes               No          No          No   

  StreamingMovies        Contract PaperlessBilling              PaymentMethod  \
0              No  Month-to-month              Yes           Electronic check   
1              No        One year               No               Mailed check   
2              No  Month-to-month              Yes               Mailed check   
3              No        One year               No  Bank transfer (automatic)   
4              No  Month-to-month              Yes           Electronic check   
5             Yes  Month-to-month              Yes           Electronic check   
6              No  Month-to-month              Yes    Credit card (automatic)   
7              No  Month-to-month               No               Mailed check   
8             Yes  Month-to-month              Yes           Electronic check   
9              No        One year               No  Bank transfer (automatic)   

   MonthlyCharges TotalCharges Churn  
0           29.85        29.85    No  
1           56.95       1889.5    No  
2           53.85       108.15   Yes  
3           42.30      1840.75    No  
4           70.70       151.65   Yes  
5           99.65        820.5   Yes  
6           89.10       1949.4    No  
7           29.75        301.9    No  
8          104.80      3046.05   Yes  
9           56.15      3487.95    No  

% S unique values OnlineBackup
No                     3088
Yes                    2429
No internet service    1526
Name: OnlineBackup, dtype: int64

% S unique values DeviceProtection
No                     3095
Yes                    2422
No internet service    1526
Name: DeviceProtection, dtype: int64

% S unique values TechSupport
No                     3473
Yes                    2044
No internet service    1526
Name: TechSupport, dtype: int64

% S unique values StreamingTV
No                     2810
Yes                    2707
No internet service    1526
Name: StreamingTV, dtype: int64

% S unique values StreamingMovies
No                     2785
Yes                    2732
No internet service    1526
Name: StreamingMovies, dtype: int64

% S unique values Contract
Month-to-month    3875
Two year          1695
One year          1473
Name: Contract, dtype: int64

% S unique values PaperlessBilling
Yes    4171
No     2872
Name: PaperlessBilling, dtype: int64

% S unique values PaymentMethod
Electronic check             2365
Mailed check                 1612
Bank transfer (automatic)    1544
Credit card (automatic)      1522
Name: PaymentMethod, dtype: int64



% S unique values Churn
No     5174
Yes    1869
Name: Churn, dtype: int64





column is %S OnlineBackup

<Figure size 432x288 with 0 Axes>

<Figure size 432x288 with 0 Axes>

<Figure size 432x288 with 0 Axes>

<Figure size 432x288 with 0 Axes>

<Figure size 432x288 with 0 Axes>

<Figure size 432x288 with 0 Axes>

<Figure size 432x288 with 0 Axes>

<Figure size 432x288 with 0 Axes>

column is %S DeviceProtection

column is %S TechSupport

column is %S StreamingTV

column is %S StreamingMovies

column is %S Contract

column is %S PaperlessBilling

column is %S PaymentMethod

column is %S Churn

Displaying tree image

[15:19:34] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:576: 
Parameters: { "eed" } might not be used.

  This could be a false alarm, with some parameters getting used by language bindings but
  then being mistakenly passed down to XGBoost core, or some parameter actually being used
  but getting flagged wrongly here. Please open an issue if you find any such cases.


[15:19:34] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.

C:\Users\HP\anaconda3\lib\site-packages\xgboost\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].
  warnings.warn(label_encoder_deprecation_msg, UserWarning)

Datatypes : Method        object
Accuracy     float64
Precision    float64
Recall       float64
F1 score     float64
ROC score    float64
dtype: object

Dataframe shape : (6, 6)

Describe        Accuracy  Precision    Recall  F1 score  ROC score
count  6.000000   6.000000  6.000000  6.000000   6.000000
mean   0.780932   0.582531  0.463123  0.512198   0.674147
std    0.030644   0.080497  0.032123  0.028070   0.016875
min    0.724627   0.449239  0.422414  0.475728   0.652048
25%    0.775195   0.557962  0.438937  0.488287   0.660039
50%    0.792051   0.600701  0.468391  0.526427   0.682185
75%    0.794535   0.607364  0.478448  0.532582   0.686323
max    0.811923   0.691244  0.508621  0.534400   0.688105

sample records                        Method  Accuracy  Precision    Recall  F1 score  \
0      DecisionTreeClassifier  0.724627   0.449239  0.508621  0.477089   
0  GradientBoostingClassifier  0.811923   0.691244  0.431034  0.530973   
0               XGBClassifier  0.794890   0.608856  0.474138  0.533118   
0        KNeighborsClassifier  0.770050   0.544444  0.422414  0.475728   
0          LogisticRegression  0.790632   0.598513  0.462644  0.521880   
0                         SVC  0.793471   0.602888  0.479885  0.534400   

   ROC score  
0   0.652048  
0   0.683943  
0   0.687116  
0   0.653243  
0   0.680426  
0   0.688105  

Max accuracy
0.8119233498935415
                       Method  Accuracy  Precision    Recall  F1 score  \
0  GradientBoostingClassifier  0.811923   0.691244  0.431034  0.530973   

   ROC score  
0   0.683943  

pickle for Gradientclassifier: accuracy: 0.8119233498935415